scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-10-07 02:59:05 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 103

Merge pull request #19323 from BerriAI/litellm_fix_stability_issues12

Browse source 
[Fix] Bedrock stability model usage issues
This commit is contained in:
Sameer Kankute 2026-01-19 19:43:32 +05:30 • committed by GitHub
commit 9e405ce6cc
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
14 changed files with 180 additions and 97 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

52
docs/my-website/docs/providers/stability.md
View file

@ -173,6 +173,14 @@ Stability AI returns images in base64 format. The response is OpenAI-compatible:
Stability AI supports various image editing operations including inpainting, upscaling, outpainting, background removal, and more.
:::info Optional Parameters
**Important:** Different Stability models have different parameter requirements:
- Some models don't require a `prompt` (e.g., upscaling, background removal)
- The `style-transfer` model uses `init_image` and `style_image` instead of `image`
- The `outpaint` model requires numeric parameters (`left`, `right`, `up`, `down`)
LiteLLM automatically handles these differences for you.
:::
### Usage - LiteLLM Python SDK
#### Inpainting (Edit with Mask)
@ -217,11 +225,11 @@ response = image_edit(
creativity=0.3, # 0-0.35, higher = more creative
)
# Fast upscaling - quick upscaling
# Fast upscaling - quick upscaling (no prompt needed)
response = image_edit(
model="stability/stable-fast-upscale-v1:0",
image=open("low_res_image.png", "rb"),
prompt="Quickly upscale this image",
# No prompt required for fast upscale
)
print(response)
```
@ -259,7 +267,7 @@ os.environ['STABILITY_API_KEY'] = "your-api-key"
response = image_edit(
model="stability/stable-image-remove-background-v1:0",
image=open("portrait.png", "rb"),
prompt="Remove the background",
# No prompt required for fast upscale
)
print(response)
```
@ -329,10 +337,29 @@ response = image_edit(
model="stability/stable-image-erase-object-v1:0",
image=open("scene.png", "rb"),
mask=open("object_mask.png", "rb"), # Mask the object to erase
prompt="Remove the object",
# No prompt needed
)
print(response)
```
#### Style Transfer
```python showLineNumbers
from litellm import image_edit
import os
os.environ['STABILITY_API_KEY'] = "your-api-key"
# Transfer style from one image to another
# Note: Uses init_image (via image param) and style_image
response = image_edit(
model="stability/stable-style-transfer-v1:0",
image=open("content_image.png", "rb"), # Maps to init_image
style_image=open("style_reference.png", "rb"), # Style to apply
fidelity=0.5, # 0-1, balance between content and style
# No prompt needed
)
print(response)
### Supported Image Edit Models
@ -419,6 +446,23 @@ response = image_edit(
)
print(response)
```
# Fast upscale without prompt
response = image_edit(
model="bedrock/stability.stable-fast-upscale-v1:0",
image=open("low_res_image.png", "rb"),
)
# Outpaint with numeric parameters
response = image_edit(
model="bedrock/stability.stable-outpaint-v1:0",
image=open("original_image.png", "rb"),
left=100, # Automatically converted to int
right=100,
up=50,
down=50,
)
print(response)
### Supported Bedrock Stability Models

6
litellm/images/main.py
View file

@ -714,8 +714,8 @@ def image_variation(
@client
def image_edit( # noqa: PLR0915
image: Union[FileTypes, List[FileTypes]],
prompt: str,
image: Optional[Union[FileTypes, List[FileTypes]]] = None,
prompt: Optional[str]= None,
model: Optional[str] = None,
mask: Optional[str] = None,
n: Optional[int] = None,
@ -766,7 +766,7 @@ def image_edit( # noqa: PLR0915
_is_async = kwargs.pop("async_call", False) is True
# add images / or return a single image
images = image if isinstance(image, list) else [image]
images = image if isinstance(image, list) else ([image] if image is not None else [])
headers_from_kwargs = kwargs.get("headers")
merged_extra_headers: Dict[str, Any] = {}

5
litellm/llms/azure_ai/image_edit/flux2_transformation.py
View file

@ -88,7 +88,7 @@ class AzureFoundryFlux2ImageEditConfig(OpenAIImageEditConfig):
self,
model: str,
prompt: Optional[str],
image: FileTypes,
image: Optional[FileTypes],
image_edit_optional_request_params: Dict,
litellm_params: GenericLiteLLMParams,
headers: dict,
@ -102,6 +102,9 @@ class AzureFoundryFlux2ImageEditConfig(OpenAIImageEditConfig):
if prompt is None:
raise ValueError("FLUX 2 image edit requires a prompt.")
if image is None:
raise ValueError("FLUX 2 image edit requires an image.")
image_b64 = self._convert_image_to_base64(image)
# Build request body with required params

2
litellm/llms/base_llm/image_edit/transformation.py
View file

@ -93,7 +93,7 @@ class BaseImageEditConfig(ABC):
self,
model: str,
prompt: Optional[str],
image: FileTypes,
image: Optional[FileTypes],
image_edit_optional_request_params: Dict,
litellm_params: GenericLiteLLMParams,
headers: dict,

12
litellm/llms/bedrock/image_edit/handler.py
View file

@ -62,7 +62,7 @@ class BedrockImageEdit(BaseAWSLLM):
self,
model: str,
image: list,
prompt: str,
prompt: Optional[str],
model_response: ImageResponse,
optional_params: dict,
logging_obj: LitellmLogging,
@ -127,7 +127,7 @@ class BedrockImageEdit(BaseAWSLLM):
timeout: Optional[Union[float, httpx.Timeout]],
model: str,
logging_obj: LitellmLogging,
prompt: str,
prompt: Optional[str],
model_response: ImageResponse,
client: Optional[AsyncHTTPHandler] = None,
) -> ImageResponse:
@ -163,7 +163,7 @@ class BedrockImageEdit(BaseAWSLLM):
self,
model: str,
image: list,
prompt: str,
prompt: Optional[str],
optional_params: dict,
api_base: Optional[str],
extra_headers: Optional[dict],
@ -176,7 +176,7 @@ class BedrockImageEdit(BaseAWSLLM):
Args:
model (str): The model to use for the image edit
image (list): The images to edit
prompt (str): The prompt for the edit
prompt (Optional[str]): The prompt for the edit
optional_params (dict): The optional parameters for the image edit
api_base (Optional[str]): The base URL for the Bedrock API
extra_headers (Optional[dict]): The extra headers to include in the request
@ -248,7 +248,7 @@ class BedrockImageEdit(BaseAWSLLM):
self,
model: str,
image: list,
prompt: str,
prompt: Optional[str],
optional_params: dict,
) -> dict:
"""
@ -276,7 +276,7 @@ class BedrockImageEdit(BaseAWSLLM):
model_response: ImageResponse,
model: str,
logging_obj: LitellmLogging,
prompt: str,
prompt: Optional[str],
response: httpx.Response,
data: dict,
) -> ImageResponse:

95
litellm/llms/bedrock/image_edit/stability_transformation.py
View file

@ -150,11 +150,11 @@ class BedrockStabilityImageEditConfig(BaseImageEditConfig):
return mapped_params
def transform_image_edit_request(
def transform_image_edit_request( #noqa: PLR0915
self,
model: str,
prompt: Optional[str],
image: FileTypes,
image: Optional[FileTypes],
image_edit_optional_request_params: Dict,
litellm_params: GenericLiteLLMParams,
headers: dict,
@ -164,32 +164,38 @@ class BedrockStabilityImageEditConfig(BaseImageEditConfig):
Returns the request body dict that will be JSON-encoded by the handler.
"""
if prompt is None:
raise ValueError("Bedrock Stability image edit requires a prompt.")
# Build Bedrock Stability request
data: Dict[str, Any] = {
"prompt": prompt,
"output_format": "png", # Default to PNG
}
# Convert image to base64
image_b64: str
if hasattr(image, 'read') and callable(getattr(image, 'read', None)):
# File-like object (e.g., BufferedReader from open())
image_bytes = image.read() # type: ignore
image_b64 = base64.b64encode(image_bytes).decode('utf-8') # type: ignore
elif isinstance(image, bytes):
# Raw bytes
image_b64 = base64.b64encode(image).decode('utf-8')
elif isinstance(image, str):
# Already a base64 string
image_b64 = image
else:
# Try to handle as bytes
image_b64 = base64.b64encode(bytes(image)).decode('utf-8') # type: ignore
# Add prompt only if provided (some models don't require it)
if prompt is not None and prompt != "":
data["prompt"] = prompt
# Convert image to base64 if provided
if image is not None:
image_b64: str
if hasattr(image, 'read') and callable(getattr(image, 'read', None)):
# File-like object (e.g., BufferedReader from open())
image_bytes = image.read() # type: ignore
image_b64 = base64.b64encode(image_bytes).decode('utf-8') # type: ignore
elif isinstance(image, bytes):
# Raw bytes
image_b64 = base64.b64encode(image).decode('utf-8')
elif isinstance(image, str):
# Already a base64 string
image_b64 = image
else:
# Try to handle as bytes
image_b64 = base64.b64encode(bytes(image)).decode('utf-8') # type: ignore
data["image"] = image_b64
# For style-transfer models, map image to init_image
model_lower = model.lower()
if "style-transfer" in model_lower:
data["init_image"] = image_b64
else:
data["image"] = image_b64
# Add optional params (already mapped in map_openai_params)
for key, value in image_edit_optional_request_params.items(): # type: ignore
@ -221,30 +227,43 @@ class BedrockStabilityImageEditConfig(BaseImageEditConfig):
file_b64 = str(file_bytes)
data[key] = file_b64
continue
# Supported text fields
if key in [
"negative_prompt",
"aspect_ratio",
"seed",
"output_format",
"model",
"mode",
# Numeric fields that need to be converted to int/float
numeric_int_fields = ["left", "right", "up", "down", "seed"]
numeric_float_fields = [
"strength",
"style_preset",
"creativity",
"control_strength",
"grow_mask",
"left",
"right",
"up",
"down",
"select_prompt",
"search_prompt",
"fidelity",
"composition_fidelity",
"style_strength",
"change_strength",
]
if key in numeric_int_fields:
# Convert to int (these are pixel values for outpaint)
try:
data[key] = int(value) # type: ignore
except (ValueError, TypeError):
data[key] = value # type: ignore
elif key in numeric_float_fields:
# Convert to float
try:
data[key] = float(value) # type: ignore
except (ValueError, TypeError):
data[key] = value # type: ignore
# Supported text fields
elif key in [
"negative_prompt",
"aspect_ratio",
"output_format",
"model",
"mode",
"style_preset",
"select_prompt",
"search_prompt",
]:
data[key] = value # type: ignore

12
litellm/llms/gemini/image_edit/transformation.py
View file

@ -81,21 +81,23 @@ class GeminiImageEditConfig(BaseImageEditConfig):
self,
model: str,
prompt: Optional[str],
image: FileTypes,
image: Optional[FileTypes],
image_edit_optional_request_params: Dict[str, Any],
litellm_params: GenericLiteLLMParams,
headers: dict,
) -> Tuple[Dict[str, Any], Optional[RequestFiles]]:
inline_parts = self._prepare_inline_image_parts(image)
inline_parts = self._prepare_inline_image_parts(image) if image else []
if not inline_parts:
raise ValueError("Gemini image edit requires at least one image.")
if prompt is None:
raise ValueError("Gemini image edit requires a prompt.")
# Build parts list with image and prompt (if provided)
parts = inline_parts.copy()
if prompt is not None and prompt != "":
parts.append({"text": prompt})
contents = [
{
"parts": inline_parts + [{"text": prompt}],
"parts": parts,
}
]

22
litellm/llms/openai/image_edit/dalle2_transformation.py
View file

@ -31,7 +31,7 @@ class DallE2ImageEditConfig(OpenAIImageEditConfig):
self,
model: str,
prompt: Optional[str],
image: FileTypes,
image: Optional[FileTypes],
image_edit_optional_request_params: Dict,
litellm_params: GenericLiteLLMParams,
headers: dict,
@ -40,18 +40,20 @@ class DallE2ImageEditConfig(OpenAIImageEditConfig):
Transform image edit request for DALL-E-2.
DALL-E-2 only accepts a single image with field name "image" (not "image[]").
"""
if prompt is None:
raise ValueError("DALL-E-2 image edit requires a prompt.")
request = ImageEditRequestParams(
model=model,
image=image,
prompt=prompt,
"""
request_params = {
"model": model,
**image_edit_optional_request_params,
)
}
if image is not None:
request_params["image"] = image
if prompt is not None:
request_params["prompt"] = prompt
request = ImageEditRequestParams(**request_params)
request_dict = cast(Dict, request)
#########################################################
# Separate images and masks as `files` and send other parameters as `data`
#########################################################

20
litellm/llms/openai/image_edit/transformation.py
View file

@ -80,7 +80,7 @@ class OpenAIImageEditConfig(BaseImageEditConfig):
self,
model: str,
prompt: Optional[str],
image: FileTypes,
image: Optional[FileTypes],
image_edit_optional_request_params: Dict,
litellm_params: GenericLiteLLMParams,
headers: dict,
@ -91,15 +91,17 @@ class OpenAIImageEditConfig(BaseImageEditConfig):
Handles multipart/form-data for images. Uses "image[]" field name
to support multiple images (e.g., for gpt-image-1).
"""
if prompt is None:
raise ValueError("OpenAI image edit requires a prompt.")
request = ImageEditRequestParams(
model=model,
image=image,
prompt=prompt,
# Build request params, only including non-None values
request_params = {
"model": model,
**image_edit_optional_request_params,
)
}
if image is not None:
request_params["image"] = image
if prompt is not None:
request_params["prompt"] = prompt
request = ImageEditRequestParams(**request_params)
request_dict = cast(Dict, request)
#########################################################

18
litellm/llms/recraft/image_edit/transformation.py
View file

@ -102,7 +102,7 @@ class RecraftImageEditConfig(BaseImageEditConfig):
self,
model: str,
prompt: Optional[str],
image: FileTypes,
image: Optional[FileTypes],
image_edit_optional_request_params: Dict,
litellm_params: GenericLiteLLMParams,
headers: dict,
@ -114,15 +114,15 @@ class RecraftImageEditConfig(BaseImageEditConfig):
https://www.recraft.ai/docs#image-to-image
"""
if prompt is None:
raise ValueError("Recraft image edit requires a prompt.")
request_body: RecraftImageEditRequestParams = RecraftImageEditRequestParams(
model=model,
prompt=prompt,
strength=image_edit_optional_request_params.pop("strength", self.DEFAULT_STRENGTH),
request_params = {
"model": model,
"strength": image_edit_optional_request_params.pop("strength", self.DEFAULT_STRENGTH),
**image_edit_optional_request_params,
)
}
if prompt is not None:
request_params["prompt"] = prompt
request_body = RecraftImageEditRequestParams(**request_params)
request_dict = cast(Dict, request_body)
#########################################################
# Reuse OpenAI logic: Separate images as `files` and send other parameters as `data`

9
litellm/llms/stability/image_edit/transformations.py
View file

@ -171,7 +171,7 @@ class StabilityImageEditConfig(BaseImageEditConfig):
self,
model: str,
prompt: Optional[str],
image: FileTypes,
image: Optional[FileTypes],
image_edit_optional_request_params: Dict,
litellm_params: GenericLiteLLMParams,
headers: dict,
@ -190,11 +190,14 @@ class StabilityImageEditConfig(BaseImageEditConfig):
}
# Add prompt only if provided (some Stability endpoints don't require it)
if prompt is not None:
if prompt is not None and prompt != "":
data["prompt"] = prompt
# Handle image parameter - could be a single file or list
image_file = image[0] if isinstance(image, list) else image # type: ignore
files: Dict[str, Any] = {"image": image_file}
files: Dict[str, Any] = {}
if image is not None:
image_file = image[0] if isinstance(image, list) else image # type: ignore
files["image"] = image_file
# Add optional params (already mapped in map_openai_params)
for key, value in image_edit_optional_request_params.items(): # type: ignore

12
litellm/llms/vertex_ai/image_edit/vertex_gemini_transformation.py
View file

@ -152,22 +152,24 @@ class VertexAIGeminiImageEditConfig(BaseImageEditConfig, VertexLLM):
self,
model: str,
prompt: Optional[str],
image: FileTypes,
image: Optional[FileTypes],
image_edit_optional_request_params: Dict[str, Any],
litellm_params: GenericLiteLLMParams,
headers: dict,
) -> Tuple[Dict[str, Any], Optional[RequestFiles]]:
inline_parts = self._prepare_inline_image_parts(image)
inline_parts = self._prepare_inline_image_parts(image) if image else []
if not inline_parts:
raise ValueError("Vertex AI Gemini image edit requires at least one image.")
if prompt is None:
raise ValueError("Vertex AI Gemini image edit requires a prompt.")
# Build parts list with image and prompt (if provided)
parts = inline_parts.copy()
if prompt is not None and prompt != "":
parts.append({"text": prompt})
# Correct format for Vertex AI Gemini image editing
contents = {
"role": "USER",
"parts": inline_parts + [{"text": prompt}]
"parts": parts
}
request_body: Dict[str, Any] = {"contents": contents}

2
litellm/llms/vertex_ai/image_edit/vertex_imagen_transformation.py
View file

@ -144,7 +144,7 @@ class VertexAIImagenImageEditConfig(BaseImageEditConfig, VertexLLM):
self,
model: str,
prompt: Optional[str],
image: FileTypes,
image: Optional[FileTypes],
image_edit_optional_request_params: Dict[str, Any],
litellm_params: GenericLiteLLMParams,
headers: dict,

10
litellm/proxy/image_endpoints/endpoints.py
View file

@ -244,8 +244,10 @@ async def image_edit_api(
if mask is None and mask_array is not None:
mask = mask_array
if image is None:
raise HTTPException(status_code=422, detail="Field required: image")
# if image is None:
# raise HTTPException(status_code=422, detail="Field required: image")
# Note: Image is optional for some models (e.g., Bedrock Stability style-transfer)
# The validation will be done at the model level if image is truly required
from litellm.proxy.proxy_server import (
_read_request_body,
@ -272,6 +274,10 @@ async def image_edit_api(
data["image"] = image_files
if mask_files:
data["mask"] = mask_files
# Ensure prompt exists in data (default to None for models that don't require it)
if "prompt" not in data:
data["prompt"] = None
data["model"] = (
model